Public/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-18 16:55:05 +02:00
Code Issues Packages Projects Releases Wiki Activity

vulkan: small M matrix optimizations for qwen (#28457)

Browse Source 
* vulkan: optimize m=1 mul_mat by swapping A/B

* vulkan: Improve small M perf

Allow split_k with small M.

Make small vs med tile selection (for coopmat2) depend on M, not just N.
This commit is contained in:
Jeff Bolz
2026-09-10 12:20:18 -05:00
committed by GitHub
parent 52d4268656
commit 6788edb4f3
2 changed files with 28 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test-backend-ops.cpp
+9 -1
View File
@@ -9683,9 +9683,17 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_MXFP4, GGML_TYPE_F32, 2880, 32, 2880, {1, 1}, {1, 1}));
// m == 1, with n on both sides of MMVF_MAX_BATCH_SIZE (8): mmvf below, operand swap above
for (int64_t n : {1, 7, 8, 9, 16, 128, 512}) {
for (int64_t n : {1, 7, 8, 9, 16, 127, 128, 511, 512}) {
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F32, GGML_TYPE_F32, 1, n, 2048, {1, 1}, {1, 1}));
}
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F16, GGML_TYPE_F32, 1, 512, 2048, {1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F16, GGML_TYPE_F16, 1, 512, 2048, {1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F16, GGML_TYPE_F32, 1, 509, 2051, {1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F16, GGML_TYPE_F16, 1, 509, 2051, {1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F32, GGML_TYPE_F32, 31, 509, 2051, {1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F32, GGML_TYPE_F32, 32, 509, 2112, {1, 1}, {1, 1}));
test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q8_0, GGML_TYPE_F32, 32, 509, 2112, {1, 1}, {1, 1}));
#if 0
{